[FFmpeg-trac] #2686(avcodec:open): Native AAC encoder collapses at high bitrates on some samples

FFmpeg trac at avcodec.org
Sun Aug 30 06:49:24 CEST 2015


#2686: Native AAC encoder collapses at high bitrates on some samples
-------------------------------------+-------------------------------------
             Reporter:  Kamedo2      |                    Owner:
                 Type:  defect       |  klaussfreire
             Priority:  normal       |                   Status:  open
              Version:  git-master   |                Component:  avcodec
             Keywords:  aac          |               Resolution:
  regression                         |               Blocked By:
             Blocking:               |  Reproduced by developer:  1
Analyzed by developer:  0            |
-------------------------------------+-------------------------------------

Comment (by Kamedo2):

 Replying to [comment:402 atomnuker]:
 > I cannot replicate this bug anymore so it's probably fixed, could you
 test with the newest git master to see if it causes problems?
 {{{
 OK ffmpeg74678-g6701c92 -y -i ffmpeg_aac_error1.flac -c:a aac -strict
 experimental -b:a 96k -ar 8000 out.mp4
 OK ffmpeg74678-g6701c92 -y -i ffmpeg_aac_error1.flac -c:a aac -strict
 experimental -q:a 1 -ar 8000 out.mp4
 OK ffmpeg74721-g4bd99f7 -y -i ffmpeg_aac_error1.flac -c:a aac -strict
 experimental -b:a 96k -ar 8000 out.mp4
 OK ffmpeg74721-g4bd99f7 -y -i ffmpeg_aac_error1.flac -c:a aac -strict
 experimental -q:a 1 -ar 8000 out.mp4
 }}}
 Yes, it was fixed.


 > I'll look into what causes the faac coder to get stuck at high bitrates.
 {{{
 OK ffmpeg74678-g6701c92 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 19000 out.mp4
 OK ffmpeg74678-g6701c92 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 17000 out.mp4
 NG ffmpeg74678-g6701c92 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 15000 out.mp4
 NG ffmpeg74678-g6701c92 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 13000 out.mp4
 NG ffmpeg74678-g6701c92 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 11000 out.mp4
 OK ffmpeg74678-g6701c92 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 9000 out.mp4
 OK ffmpeg74721-g4bd99f7 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 19000 out.mp4
 OK ffmpeg74721-g4bd99f7 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 17000 out.mp4
 NG ffmpeg74721-g4bd99f7 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 15000 out.mp4
 NG ffmpeg74721-g4bd99f7 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 13000 out.mp4
 NG ffmpeg74721-g4bd99f7 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 11000 out.mp4
 OK ffmpeg74721-g4bd99f7 -y -i ffmpeg_aac_lead_voice.flac -c:a aac -strict
 experimental -b:a 320k -cutoff 9000 out.mp4
 }}}
 The bug still exists in the git master. It induces infinite loop when the
 cutoff is 15, 13, 11kHz, but not 9kHz. When the cutoff is 7kHz, the
 initial setup takes 16 seconds, which is unnaturally slow, but the encoded
 sound is OK. ffmpeg_aac_lead_voice.flac is 44.1kHz '''mono'''.

--
Ticket URL: <https://trac.ffmpeg.org/ticket/2686#comment:403>
FFmpeg <https://ffmpeg.org>
FFmpeg issue tracker


More information about the FFmpeg-trac mailing list